Police corruption and crime: Evidence from Africa

Abstract

Using data from the Afrobarometer surveys, this paper finds that people living in regions in which police corruption is more prevalent are more likely to report that they or someone in their family have been victims of physical assault. People living in more corrupted regions are also more likely to report that they or someone in their family has had something stolen from their home. We find no statistically significant gender differences in the average marginal effects. Controlling for the incidence of corruption in other domains reduces the size of the estimated association but does not render it insignificant in terms of statistical significance or magnitude. Non-police corruption is also strongly associated with an increased risk of crime. For both types of crime, the evidence points to “transactional” police corruption (having to pay bribes to get help) rather than “predatory” police corruption (having to pay bribes to avoid problems) as driving the relationship. Finally, we show that, controlling for whether the respondent reports being a victim of either type of crime, police corruption predicts an increase in the probability that the respondent reports feeling unsafe while walking in their own neighborhood thus imposing a cost even on those who have not been victims.